Prediction of Employee attrition using Machine Learning Algorithm

Abstract

Making decisions can play a big part in administration and can even point to the crucial element in the planning process. In order to retain talented workers, management must effectively handle the well-known issue of employee attrition. It's exciting to know that ML algorithms can accurately forecast staff attrition. The aim of this research was to identify the key variables affecting employee turnover and to forecast employee attrition with a high gradation of accuracy using machine and deep learning models. The Kaggle Depository provided the dataset that was used in this investigation. The dataset, which includes 24 attributes from 1,470 employees, was produced by IBM Analytics. In order to optimize the prediction accuracy of employee attrition, the dataset was preprocessed, balanced, and divided into three distinct sets: the train, valid, and test datasets. A number of tests were conducted to demonstrate the applicability of this research. This research used JASP software to analyze the data. The best machine learning model archived f1-scores of 82.6%, precision 81.8%, recall 84.7% and accuracy 84.6% for the prediction of employee attrition.